Method for the estimation of genetic merit of animals with uncertain paternity under Bayesian inference

Abstract

The use of controlled mating or artificial insemination is impracticable in the case of large herds, mainly because of labour costs and the need to delimit areas during the breeding period. However, the exclusion of information from animals with uncertain paternity reduces genetic progress. The objectives of this study were as follows: (i) propose an iterative empirical Bayesian procedure to implement the hierarchical animal model (ITER); (ii) calculate the posterior probabilities of paternity by the maximum likelihood method following the concepts; (iii) compare an average numerator relationship matrix (ANRM), Bayesian hierarchical (HIER) models and ITER. Records of Nellore animals born between 1984 and 2006 from the zootechnical archive of Agropecuária Jacarezinho Ltda were used. For data consistency, records of contemporary groups (CGs) with fewer than three animals and animals whose records were 3.5 standard deviations above or below the mean of their CG were eliminated. After editing the data, 62,212 animals in the file and 12,876 animals in pedigree file were maintained, respectively. Spearman and Pearson correlations between the posterior mean of the genetic effects of animals were calculated to compare the ranking of animals for selection. Simulated data were used to confirm the veracity of the model. The correlations between ITER and HIER and between ITER and ANRM were similar evaluating different files, which decreased at the same proportion when only high-ranked animals were evaluated. In conclusion, the model proposed herein is a suitable computational alternative to improve the prediction of breeding values of animals in genetic evaluations using large databases, including animals with uncertain paternity.